Future-Ready Pharma
The Transformative Role of GenAI
Tarun Mathur, Chief Technology Officer, Indegene
With pharma companies ramping up their GenAI-based experiments, how can commercial, medical and clinical leaders in life sciences organizations make the right GenAI bets? In an interview with Pharma Focus America, Tarun Mathur, CTO, Indegene, sheds light on why GenAI is here to stay, and areas with maximum business impact.
1. Can you explain how GenAI is driving significant advancements in drug discovery and the specific technologies or methodologies that have been most effective in this area?
GenAI is revolutionizing the drug discovery process - which traditionally takes years and has a high attrition rate, given only 10% of candidate molecules advance to clinical trials, per industry research. We have seen GenAI add value across multiple stages of the discovery process including target identification and validation to drug interaction prediction and lead optimization. A particularly interesting development is the use of Deep Generative Models (DGM) to optimize chemical structures based on target-derived 3D models. With large, high-quality training datasets, DGMs can generate a diverse array of valid small and macromolecules, accelerating the discovery process.
2. In what ways can GenAI contribute to an increase in productivity and efficiency within the life sciences sector, particularly over the next five years?
GenAI's capacity to process large and diverse data sets and generate preliminary reports has increased its adoption across the life sciences industry. In drug discovery, for example, GenAI improves efficiency by employing predictive and classification models to generate new molecules. Beyond discovery, GenAI's impact extends across the entire drug development value chain, notably in commercial, medical, and clinical functions. Over the next five years, we anticipate significant advancements in automation and personalization within these areas, driving a substantial increase in productivity and efficiency.
3. How does GenAI enhance the drug development process, and what are the key factors that enable it to make more accurate predictions of molecular behavior?
GenAI enhances the drug development process through its unique approach to integrating domain knowledge with AI models. Its predictive capabilities rest on three core pillars: big data, advanced algorithms, and sophisticated feature engineering. By leveraging large, domain-specific datasets - both structured and unstructured - GenAI models can learn intricate molecular patterns and relationships. Utilizing deep neural networks and embeddings to capture critical molecular features, these models produce highly reliable predictions of molecular behavior. That said, iterative refinement with real-world data and expert validation is essential to ensure the accuracy and robustness of these predictions.
4. Discuss the impact of GenAI in reducing the time-to-market for new treatments. What specific stages of drug development benefit the most from this technology?
GenAI can significantly reduce the time-to-market for new treatments by streamlining processes across the drug development value chain. Large language and multimodal models, which process a diverse array of data including unstructured text, images, patient information, and omics data, play a pivotal role especially when implemented in agentic workflows. The most transformative impact of GenAI is seen in the commercial and clinical development stages. In clinical development, GenAI accelerates the creation of dossiers and regulatory documents, while in commercial operations, it enables the rapid generation of personalized content for healthcare professionals and other stakeholders.
5. How are life sciences companies leveraging GenAI to offer highly personalized treatment plans, and what has been the observed impact on patient outcomes?
We are seeing an emergent use case of GenAI impacting personalized treatment plans leading to a direct effect on improving patient outcomes. At its core, GenAI enhances patient access to tailored care by increasing the efficiency of patient support programs (PSPs). These programs often involve patient education, financial assistance, disease management tools, and various administrative tasks. The impact is particularly noticeable in specialty medicines and access to rare and orphan drugs, where benefits verification and prior authorization are traditionally time-consuming and labor-intensive. GenAI's ability to process text, audio, and images enables the automation of these processes, significantly reducing wait times for patients requiring urgent treatment. Additionally, while early in development, the ability to have personal GenAI-based agents that have access to key information enables the possibility of near real-time guidance and interventions from this technology. This has the potential to make a major impact on personalized wellness and treatment plans on an individual basis.
6. In terms of reducing healthcare costs, what role does GenAI play in the personalization of patient care, and how significant are the cost savings?
GenAI can play a vital role in reducing healthcare costs by enhancing personalized patient care and streamlining operational processes across providers, payers, and manufacturers. By automating tasks such as patient scheduling, medical coding, benefits verification, and prior authorization, analysts anticipate GenAI could deliver immediate cost savings of up to 10%, with potential reductions of up to 50% in the long term. Particularly within patient support programs (PSPs), the automation of labor-intensive tasks and the generation of actionable insights is expected to lead to cost savings of up to 75%, minimizing the need for human intervention while improving service efficiency.
7. What are the major challenges faced by life sciences companies in integrating GenAI into their existing drug development pipelines, and how are these challenges being addressed?
We see several challenges regularly flagged, primarily related to data and IT readiness, regulatory compliance, and the explainability of AI-generated insights. Many life sciences companies are actively upgrading their data and IT infrastructure and partnerships with hyperscalers to support the computational demands of GenAI applications.
Regulatory compliance poses a challenge due to GenAI’s potential for generating hallucinated responses. To address this, companies are adopting a three-pronged approach: architecting GenAI-based systems with fit-for-purpose agents and risk-mitigating workflows, incorporating a human-in-the-loop mechanism to review content and eliminate bias and inaccuracies, and implementing quality control software that flags deviations and potential hallucinations. The issue of explainability is another significant hurdle. Organizations are working to increase transparency around what GenAI models can and cannot do, while also developing rules-based mechanisms and algorithms that trace the origin and logic of AI-generated information.
8. Can you provide examples of specific cases where GenAI has led to breakthroughs in drug discovery or development, and what were the critical success factors in these cases?
Google DeepMind has had some major developments specifically targeting drug discovery. An example is a new fine-tuned model called Tx-LLM. This model is specifically tuned to solve many of the complex, data-intensive tasks involved in the development pipeline. Earlier this year, a research paper on this model was shared and it demonstrated that the model was able to outperform the existing state-of-the-art approaches for around 33% of the tasks involved. This will have a significant impact on the cost and timelines involved during the development process.
Critical to this is the careful fine-tuning of the model and the fact that the model is specifically targeting this application. DeepMind partnered with a deep domain expertise organization and curated data for this use case and developed a testing approach just for this.
As the GenAI landscape evolves, we expect to see a growing ecosystem of highly specialized models that are tuned and managed by deep domain experts.
9. How do life sciences companies ensure the accuracy and reliability of GenAI predictions in drug development, and what methodologies are used to validate these predictions?
We have seen several approaches to addressing this challenge. A human-in-the-loop approach is one of the most effective strategies to achieve this. By integrating human oversight at various stages - from design through to commercialization - companies can validate GenAI outputs, ensuring they meet the highest standards of accuracy and reliability. This approach is particularly crucial in use cases such as regulatory report automation, drug discovery predictions, and clinical trial design, where precision is paramount.
Another approach that is emerging - but rapidly growing - is using AI Feedback. We have seen that GenAI models can be specifically tuned to critique and evaluate the output of other models. This AI feedback loop can then deliver feedback to the creator models. Some advanced systems could even include both human-in-the-loop and AI feedback mechanisms.
10. Discuss the role of GenAI in identifying potential side effects and adverse reactions earlier in the drug development process. How does this improve overall patient safety?
GenAI can play an important role in identifying potential side effects and adverse reactions earlier in the drug development process, thereby enhancing overall patient safety. With the proliferation of digital channels, patients and healthcare stakeholders now have multiple avenues to report adverse events. GenAI excels at processing unstructured data from diverse sources such as social media, call centers, audio files, and images. It consolidates and identifies adverse events efficiently, streamlining the quality assurance and validation processes. This not only saves time but also enables a more granular and comprehensive search across all available data sources, ensuring that adverse events are detected and managed promptly.
11. What are the strategic priorities for life sciences companies in implementing GenAI technologies, and how do these priorities align with broader industry trends?
We see life sciences companies aligning their priorities closely with broader industry trends, particularly the need to reduce costs and keep pace with digital transformation. With nearly 50% of large global biopharma companies exceeding industry averages in SG&A expenses in 2023, cost reduction has become a critical focus, especially as the industry faces a looming patent cliff and new legislative pressures. Additionally, the evolving technology landscape and the shift toward digital content consumption by healthcare stakeholders, including HCPs, have necessitated a more omnichannel approach to commercialization. GenAI is well-suited to support these strategies, offering optimized, cost-effective outreach solutions that meet the demands of a digital-first market.
12. How does GenAI facilitate collaboration between different stakeholders in the life sciences sector, such as researchers, clinicians, and regulatory bodies?
With GenAI impacting the entire healthcare ecosystem and publicly available use cases like ChatGPT, there has been a general awareness among healthcare stakeholders about the disruptive potential of the technology. GenAI, in many ways, has the potential to strengthen collaboration due to the very nature of its technology. Being a unified platform for accepting virtually all forms of data (structured, unstructured, language, images, video) allows researchers, clinicians and life sciences companies to analyze and assess data at the same table, almost simultaneously - creating a unified collaboration ecosystem. In addition, GenAI’s ability to accept information in natural language allows various stakeholders to interact in plain language, that is understood by clinicians, technologists and regulatory experts alike. The FDA too has taken steps in this regard by notifying a framework for AI in drug manufacturing. This not only shows regulatory bodies’ inclination to adopt transformational technologies but also their willingness to get manufacturers to participate in this transformation in a safe and regulated environment.
13. In what ways is GenAI expected to transform personalized medicine in the coming years, and what are the key innovations driving this transformation?
GenAI combined with other technologies such as digital health tools and wearable devices, is helping to redefine personalized medicine. Highly capable models that can interpret and reason over complex personal information are one of the major continuous innovation areas that are fueling this transformation. Additionally, new algorithms that are making these models more efficient and can run on personal devices and protected environments with lower inference costs, will dramatically impact adoption - and ultimately transformation. We are already seeing the emergence of highly capable models that can run on smartphones and we expect to see these models continue to evolve in their advanced reasoning capabilities. This type of technological innovation will enable new patient-centric use cases at scale and transform personalized medicine.
14. What measures are being taken by life sciences companies to manage the ethical and regulatory considerations associated with the use of GenAI in drug development and patient care?
As companies evaluate their digital maturity and readiness to adopt AI, they are increasingly recognizing the need for a robust AI governance framework. In response, many pharma companies are establishing roles dedicated to AI compliance and governance, implementing processes such as AI risk management and third-party AI risk assessment.
Organizations are also embedding principles of ethical AI into their broader corporate Codes of Ethics, with oversight at the highest levels. To ensure these ethical standards are met, companies are also adopting quality control technologies that monitor and flag issues like hallucinations and bias in GenAI models, thereby maintaining the integrity and reliability of AI-generated insights.
Enterprises are increasingly setting up centralized governance teams and processes that not only establish and audit GenAI usage for ethics and compliance, but also establish the parameters for GenAI to reflect the voice of the brand(s), rules for explainability, auditing, and safety. To support this, many enterprises are establishing enterprise-wide technology enablers that include GenAI-based agents, testing and evaluation tools, and federated enterprise data catalogs that are GenAI-focused.
Author Bio
Tarun has close to three decades of experience. He has been associated with Indegene for nearly two decades. He leads the Technology domain at Indegene and his responsibilities include the development of technology-based solutions focused exclusively on the healthcare industry. His strengths lie in his technological expertise and business acumen, which help in developing various platforms and next-generation tech solutions.
